Acknowledgements

Generation of Context-SpecificePCRs using Domain-Specific

Modeling

MOTHIS WorkshopNashville, TN, Sept 30, 2007

Rohit Shenvi([email protected])

Dept. of Computer and Information SciencesAdvNet Project, HI, Dept. of HSA

University of Alabama at Birmingham

2

Advisors: Helmuth Orthner, PhD, Professor of Health Informatics, Department

of Health Services Administration, UAB Jeff Gray, PhD, Associate Professor, Department of Computer and

Information Sciences, UAB Collaborators:

Giovanni Mazza, MSHI, MSCS, System Manager, AdvNet Project, Health Informatics, Dept. of Health Services Admin., UAB

Devashish Saini, MD, MSHI, Resident Physician, University of Missouri-Columbia

Marcie Battles, MSEE, AdvNet Project Manager, HI, Dept. HSA Support:

This project has been funded by Federal funds from the National Library of Medicine, NIH, under Contract No. N01-LM-3-3513 and the National Science Foundation, under CAREER grant CCF-0643725.

Acknowledgements

3

EMS Environment and Workflow Electronic Patient Care Report (ePCR) Project

ePCR Development: Modular Approach CAB, CSLA, WCF, WPF

ePCR Model Design: GME Domain-Specific Modeling in EMS

Summary Lessons Learned

Overview of Presentation

4

9-1-1 Dispatcher Ambulance dispatched EMT Assesses, Treats & Transports Patient

Treatment

EMS Patient Flow

Patient Triage & Transport Hospital ED

Patient Care Report (PCR)

EMT Assesses, Treats & Transports Patient

6

Issues and Problems Most EMTs have moderate typing skills

Little or moderate experience with computers Concerned about extra work

“Paper + Computer = Slower Work” EMS (paper) documentation is a problem

Over 60% of required data elements are missing (Mandar Gori's Thesis Project)

Adherence to clinical protocols is low Could be a documentation problem Contact is with Online Medical Control is

avoided even when required (Dr. Devashish Saini’s MSHI Thesis)

7

ePCR Project Data acquisition using ePCR

User Friendly Interfaces to Enter Data with Minimum Actions

Implementation of Security Mechanisms such as Encryption and Authorized Access to Patient Data

Sharing data collected in the field Collect National EMS Information System

(NEMSIS)-compliant data Web Services to push/pull patient data

to/from ePCR terminal Transmitting Data to the Central Station

Paper vs. Electronic

ePCR

PCR

9

Initial Prototype: Burns

10

Initial Lessons Learned Developing new User Interfaces is very

labor intensive Maintenance and Scalability difficult

because business logic is intertwined throughout the application

EMS agencies require customization of the ePCR GUI and Business layer

Usability Study of ePCR UI requires quick development of various design alternatives

2nd Prototype: Modular Approach

11

InitialApproach

Modular Approach

2nd Prototype: Modular Approach

Layered Architecture Presentation User Interface Business Logic

Data Validation: Range Checking

Data Manipulation Conversion from common

units to international units Data Layer

ADO.NET Data Storage

12

13

Domain-Specific Modeling

Identify entities in the domain that need to be modeled

Design the metamodel

Build model interpreter

Design the domain-specific model

1414

Domain-Specific Modeling

15

Domain-Specific Modeling Promises

Generation of platform specific systems

Reduced turnaround time

Improves user-developer interaction during system development

16

Domain-Specific Modeling in EMS Business layer

needs to adapt for new requirements

Patient category (pediatric, adult, geriatric, etc.)Medical devices (pulse ox, EKG, etc.)

Drug-drug interaction engine integration with the ePCR

UI automation

17

Domain-Specific Modeling in EMS

Drug interactions Needs to be integrated within

the ePCR Domain-specific modeling to

generate wrapper code on the drug interaction API for the ePCR

Our focus Generation of business and

data layer using domain- specific modeling

18

ePCR Model Design: GME

ePCR metamodel

ePCR MandatoryBO OptionalBO PropertyOf_Mandatory PropertyOf_Optional MandatoryConn OptCon

19

ePCR Model Design: GME

ePCR domain-specific model

ePCR Context Demographics Vitals Assessment Treatment Narrative Attachments Billing Triage

20

Model to Code Transformation

Transformation logic Each object transforms

to a CSLA business class

Each objects’ properties are converted to C# code

It is difficult to implement new business rules and modify existing ones without affecting the modules where they are found

Need a mechanism to inject rules into objects at run time

The specification of the rules should be done in a simple unambiguous language and translated later to executable code

21

Lessons Learned

Limitations and Future Work Limitations of Metamodel

Initial attempt at building the metamodel. It does not contain the constructs for complex business objects

Limitations of Interpreter Functionality Not implemented to generate full fledged

CSLA business objects

Experimental Validations Formal verification and validation not done

We discussed issues related to ePCR development Deficiencies and improvements

We introduced the modeling paradigm Areas potentially benefiting by domain-

specific modeling languages

A partial implementation of the ePCR business layer was illustrated

23

Conclusions

Generation of Context-SpecificePCRs using Domain-Specific

Modeling

Questions & Thank YouRohit Shenvi

([email protected])Dept. of Computer and Information Sciences

AdvNet Project, HI, Dept. of HSA University of Alabama at Birmingham

25

Model to Code Transformation

26

Initial Prototype: Vitals

UI design using WPF and CAB WPF: Windows Presentation Foundation CAB: Composite UI Application Block

Business and data layer using CSLA CSLA: Component-Based Scalable

Logical Architecture Communication layer using WCF

WCF: Windows Communication Foundation

27

Methodology

Motivation

Modeling paradigm

Promises

Potential areas of implementation in the ePCR development cycle

28

Domain-Specific Modeling in EMS